Adata’s XPG SX8200 Pro 1 TB, reviewed

Last year, we reviewed an NVMe-equipped Adata SSD for the very first time. The XPG SX8200 came in guns blazing, and its solid performance and modest price tag earned it our official endorsement. But now it looks like we’ll have to table that recommendation, because Adata has whipped up a sequel, the XPG SX8200 Pro.

Adata XPG SX8200 Pro
Capacity Max sequential (MB/s) Max random (IOps)
Read Write Read Write
256 GB 3500 1200 220K 290K
512 GB 3500 2300 390K 380K
1 TB 3500 3000 390K 380K

Just like the original SX8200, the Pro is an NVMe-enabled, M.2, four-lane-PCIe drive, with 64-layer Micron TLC and a Silicon Motion controller humming along under the hood. But despite all those similarities, Adata’s product pages claim a variety of significant speed improvements. Random read and write IOps have seen hefty gains, and the 1 TB version’s sequential write speeds have skyrocketed from 1700 MB/s to 3000 MB/s. What accounts for the difference? Maybe it’s the XPG sticker that Adata now ships in the box.

A quick squint at the bare drive reveals that the brains of the operation is Silicon Motion’s SM2262EN controller, a new-and-improved flavor of the SM2262 that underpinned the SX8200. Silicon Motion’s spec sheet reveals precious few differences between the old and the new controllers—they’re both eight-channel affairs with the same interfaces and capabilities—but the SM2262EN’s maximum performance ratings are higher. Unfortunately, the SX8200 Pro sample we received is the 1 TB version, so we won’t be able to do a true and direct comparison of the two controllers, since our original SX8200 review unit was a 480 GB drive. 

The controller really does seem to be the only major difference between the old and new drives. They both lean on pseudo-SLC caching to reach their peak speeds, both omit hardware encryption acceleration, both have the same five-year warranty, and both are rated for the same number of total bytes written—640 TB for the 1 TB flavor. One thing that has drastically changed is the price: The 1-TB SX8200 Pro is available at at Amazon for $193, which is a far cry from the $190 that 480 GB version of the prior SX8200 commanded at the time we reviewed it.

So let’s see what a new controller and a fresh coat of sticker does for the SX8200 line. To the bench!

 

IOMeter — Sequential and random performance

IOMeter fuels much of our latest storage test suite, including our sequential and random I/O tests. These tests are run across the full capacity of the drive at two queue depths. The QD1 tests simulate a single thread, while the QD4 results emulate a more demanding desktop workload. For perspective, 87% of the requests in our old DriveBench 2.0 trace of real-world desktop activity have a queue depth of four or less. Clicking the buttons below the graphs switches between results charted at the different queue depths. Our sequential tests use a relatively large 128-KB block size.



The SX8200 Pro 1 TB’s read speeds aren’t markedly different from the SX8200 500 GB’s, but the writes show tremendous improvement. The Pro doubles the write rates of our older SX8200 review unit. Of course, some of that improvement is simply due to the capacity difference, but nonetheless we’re impressed. Still, Samsung’s 970 EVO 1 TB enjoys a commanding lead in the sequential write tests. TurboWrite seems to remain the gold standard for pseudo-SLC caching speed. Let’s have a look at random rates.



Adata’s claims of random-rate gains do not manifest at QD1, but the Pro ekes out some minor wins over the SX8200 in our QD4 tests.

Our initial IOMeter tests seem to corroborate Adata’s story of how much better the SX8200 Pro is versus its predecessor. But we’ve got a lot more tests to pit against the newcomer.

 

Sustained and scaling I/O rates

Our sustained IOMeter test hammers drives with 4-KB random writes for 30 minutes straight. It uses a queue depth of 32, a setting that should result in higher speeds that saturate each drive’s overprovisioned area more quickly. This lengthy—and heavy—workload isn’t indicative of typical PC use, but it provides a sense of how the drives react when they’re pushed to the brink.

We’re reporting IOPS rather than response times for these tests. Click the buttons below the graph to switch between SSDs.


The SX8200 Pro’s peak random write rates hover around the same range as the original SX8200’s. But while the SX8200  hit that speed only during a single, fleeting burst, the SX8200 Pro managed to hold to its highest speeds over the span of minutes. The drives both decline to a similar steady-state rate between 20K and 30K IOps.

The absolute peak figure masks the difference in behavior between the two drives that we observed during the high-speed phase. Silicon Motion’s firmware for the SM2262EN clearly has been tuned to milk its pseudo-SLC caching potential for all it’s worth. But as we noted before, Samsung has some secrets that Adata has yet to penetrate; the 970 EVO 1 TB’s steady-state write rates are almost a full 50% faster than that of the XPG SX8200 Pro 1 TB.

Our final IOMeter test examines performance scaling across a broad range of queue depths. We ramp all the way up to a queue depth of 128. Don’t expect AHCI-based drives to scale past 32, though—that’s the maximum depth of their native command queues.

For this test, we use a database access pattern comprising 66% reads and 33% writes, all of which are random. The test runs after 30 minutes of continuous random writes that put the drive in a simulated used state. Click the buttons below the graph to switch between the different drives. Note that each drive uses a different scale for IOPS to allow us to view the shape of its curves.


Adata’s new drive offers smooth scaling until a queue depth of about 32 before leveling off. Not bad for a consumer-oriented drive.


When viewed side by side, it seems that the SX8200 Pro 1 TB does slightly worse in this test than its 480-GB predecessor. Perhaps the firmware’s new burst-speed grunt comes at the cost of performance in more demanding workloads.

Nevertheless, the SX8200 Pro is lookin’ good. IOMeter synthetics have proved troublesome in some of our previous Adata reviews, but this drive hasn’t really broken a sweat yet. Now it’s time to see what its real-world performance is like.

 

TR RoboBench — Real-world transfers

RoboBench trades synthetic tests with random data for real-world transfers with a range of file types. Developed by our in-house coder, Bruno “morphine” Ferreira, this benchmark relies on the multi-threaded robocopy command build into Windows. We copy files to and from a wicked-fast RAM disk to measure read and write performance. We also cut the RAM disk out of the loop for a copy test that transfers the files to a different location on the SSD.

Robocopy uses eight threads by default, and we’ve also run it with a single thread. Our results are split between two file sets, whose vital statistics are detailed below. The compressibility percentage is based on the size of the file set after it’s been crunched by 7-Zip.

Number of files Average file size Total size Compressibility
Media 459 21.4 MB 9.58 GB 0.8%
Work 84,652 48.0 KB 3.87 GB 59%

RoboBench’s write and copy tests run after the drives have been put into a simulated used state with 30 minutes of 4-KB random writes. The pre-conditioning process is scripted, as is the rest of the test, ensuring that drives have the same amount of time to recover.

The media set is made up of large movie files, high-bitrate MP3s, and 18-MP RAW and JPG images. There are only a few hundred files in total, and the data set isn’t amenable to compression. The work set comprises loads of TR files, including documents, spreadsheets, and web-optimized images. It also includes a stack of programming-related files associated with our old Mozilla compiling test and the Visual Studio test on the next page. The average file size is measured in kilobytes rather than megabytes, and the files are mostly compressible.

Let’s take a look at the media set first. The buttons switch between read, write, and copy results.



The SX8200 Pro actually trails its predecessor slightly in the write tests, but more than makes up for it in the copy tests. It’s the fastest reader and copier yet in the media set with a single thread. The write performance gap between the XPG and Samsung’s 970 EVO 1 TB seems to have shrunk substantially in the real world. What about the more difficult work set?



This time it snags the records across all three tests at 1T. Adata’s new drive is a file-pushing monster. 

The SX8200 Pro doesn’t always manage to outdo the original SX8200, but that was a high bar to begin with. The Pro put on a fine showing in RoboBench, about on par (in the aggregate) with Samsung’s perennially potent performers. Our next and last set of tests will measure the drive’s abilities as a primary boot drive.

 

Boot times

Until now, all of our tests have been conducted with the SSDs connected as secondary storage. This next batch uses them as system drives.

We’ll start with boot times measured two ways. The bare test depicts the time between hitting the power button and reaching the Windows desktop, while the loaded test adds the time needed to load four applications—Avidemux, LibreOffice, GIMP, and Visual Studio—automatically from the startup folder. These tests cover the entire boot process, including drive initialization.

Yet another unremarkable boot performance. Don’t get me wrong—unremarkable is desirable in this context. Very few SSDs tend to set themselves apart in boot-time tests, and even when they do, it’s not often in a good way.

Load times

Next, we’ll tackle load times with two sets of tests. The first group focuses on the time required to load larger files in a collection of desktop applications. We open a 790-MB 4K video in Avidemux, a 30-MB spreadsheet in LibreOffice, and a 523-MB image file in GIMP. In the Visual Studio Express test, we open a 159-MB project containing source code for Microsoft’s PowerShell.

Load times for the first three programs are recorded using PassMark AppTimer. AppTimer’s load completion detection doesn’t play nice with Visual Studio, so we’re still using a stopwatch for that one.

Similarly, these tests don’t typically do much to differentiate SSDs, but this time the SX8200 Pro snags three top finishes. Adata’s onto something here. Finally, let’s load up some games.

The SX8200 Pro is a top-tier gamer, but so is just about any SSD. To give credit where it’s due, though, the drive does net another pair of first-place trophies.

We’ve exhausted our supply of tests but have failed to exhaust the XPG SX8200 Pro. It withstood every sort of challenge we could conceive. The penultimate page of this review documents our storage testing methods, and the conclusion follows thereafter.

 

Test notes and methods

Here are the essential details for all the drives we tested:

Interface Flash controller NAND
Adata XPG SX8200 480 GB PCIe Gen3 x4 Silicon Motion SM2262 64-layer Micron 3D TLC
Adata XPG SX8200 Pro 1 TB PCIe Gen3 x4 Silicon Motion SM2262EN 64-layer Micron 3D TLC
Crucial MX500 500 GB SATA 6Gbps Silicon Motion SM2258 64-layer Micron 3D TLC
Crucial P1 500 GB PCIe Gen3 x4 Silicon Motion SM2263 64-layer Micron 3D QLC
Intel X25-M G2 160 GB SATA 3Gbps Intel PC29AS21BA0 34-nm Intel MLC
Samsung 850 EVO 1 TB SATA 6Gbps Samsung MEX 32-layer Samsung TLC
Samsung 860 EVO 1 TB SATA 6Gbps Samsung MJX 64-layer Samsung TLC
Samsung 860 QVO 1 TB SATA 6Gbps Samsung MJX 64-layer Samsung QLC
Samsung 970 EVO 1 TB PCIe Gen3 x4 Samsung Phoenix 64-layer Samsung TLC
Samsung 970 EVO Plus 1 TB PCIe Gen3 x4 Samsung Phoenix 90+-layer Samsung TLC
Toshiba RC100 PCIe Gen3 x2 Toshiba 64-layer Toshiba BiCS TLC

The SATA SSDs were connected to the motherboard’s Z270 chipset. The PCIe drives were connected via one of the motherboard’s M.2 slots, which also draw their lanes from the Z270 chipset.

We used the following system for testing:

Processor Intel Core i7-6700K
Motherboard Gigabyte Aorus Z270X-Gaming 5
Firmware F10B
Memory size 16 GB (2 DIMMs)
Memory type Corsair Vengeance LPX DDR4 at 2133 MT/s
Memory timings 15-17-17-35
System drive Corsair Force LS 240 GB with S8FM07.9 firmware
Power supply Rosewill Fortress 550 W
Operating system Windows 10 x64 1803

Thanks to Gigabyte for providing the system’s motherboard, to Intel for the CPU, to Corsair for the memory and system drive, and to Rosewill for the PSU. And thanks to the drive makers for supplying the rest of the SSDs.

We used the following versions of our test applications:

Some further notes on our test methods:

  • To ensure consistent and repeatable results, the SSDs were secure-erased before every component of our test suite. For the IOMeter database, RoboBench write, and RoboBench copy tests, the drives were put in a simulated used state that better exposes long-term performance characteristics. Those tests are all scripted, ensuring an even playing field that gives the drives the same amount of time to recover from the initial used state.

  • We run virtually all of our tests three times and report the median of the results. We run our sustained IOMeter test a second time to verify the results of the first test, and  we perform additional runs only if necessary. The sustained test runs for 30 minutes continuously, so it already samples performance over a long period.

  • Steps have been taken to ensure the CPU’s power-saving features don’t taint any of our results. All of the CPU’s low-power states have been disabled, effectively pegging the frequency at 4.0 GHz. Transitioning between power states can affect the performance of storage benchmarks, especially when dealing with short burst transfers.

The test systems’ Windows desktop was set at 1920 x 1200 at 60 Hz. Most of the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit up our forums to talk with us about them.

 

Conclusions

Adata’s XPG SX8200 was one of the quicker drives in our current standings when it first hit our storage labs, and the Pro edition is even faster. The SX8200 Pro 1 TB must certainly be among the fastest client NVMe drives we’ve ever tested. We distill the overall performance rating using an older SATA SSD as a baseline. To compare each drive, we then take the geometric mean of a basket of results from our test suite. Only drives that have been through the entire current test suite on our current rig are represented.

The XPG SX8200 Pro was just a few hairs short of snagging that coveted number-one position. Adata had more first-place finishes throughout the test suite, but the 970 EVO 1 TB’s margin of victory was enormous in a few of the IOMeter synthetics. But don’t fret, SX8200 Pro;  the 970 EVO is fantastic company to keep, even if it is the drive of yesteryear. 

To wrap up our discussion of the XPG SX8200 Pro and render our final verdict, we must examine the latest developments in the SSD price wars.  In the plots below, the most compelling position is toward the upper left corner, where the price per gigabyte is low and performance is high.

After months of endless freefall, it seems that SSD prices have somewhat stabilized. They’re still quite low, thankfully, but they have plummeted no further since our last check-in. The XPG SX8200 1 TB fetches $193 at Amazon or directly from Adata, which translates to a mere nineteen cents per gigabyte. It doesn’t feel like that long since the days when 1-TB SATA drives were around the three-hundred-dollar mark, so shelling out less than $200 for a terabyte of wicked-fast NVMe goodness feels like a steal.

Adata XPG SX8200 Pro 1 TB

January 2019

As far as the competition is concerned, well, you can pay a bit more for a 970 EVO or 970 EVO Plus and get a bit more performance, but it pays to keep in mind what sort of gains you’re ponying up for. We haven’t yet had a crack at the 970 EVO Plus 1 TB, but the vanilla 970 EVO 1 TB’s advantages primarily lay in our IOMeter synthetic tests for sequential writes and long-sustained random writes. The speed gap shrank to near nil in our real-world RoboBench and load time tests.

So if you’re a salt-of-earth sort of gerbil with ordinary storage needs, the XPG SX8200 Pro 1 TB will give you everything you need and more. Its competent handling of our IOMeter tests and blazing speeds in our file-transfer and stopwatch tests left us with the warm fuzzies. Its remarkably reasonable asking price was just icing on the cake. Adata has put out a real winner in the SX8200 Pro.

Comments closed
    • ronch
    • 4 months ago

    More importantly I hope SSD prices don’t shoot through the roof soon. Me thinks them SSD makers are quietly hatching a plan to raise prices again. After all, a lot of NAND producers are the same good old DRAM makers: Samsung, Micron, Toshiba. Corrections are welcome.

    • Waco
    • 5 months ago

    Bumping this – Amazon has a $15 off coupon for this drive. $167 shipped.

    • cynan
    • 6 months ago

    Definitely a win write-wise (at least for the first couple of minutes) over the sx8200 with the vanilla controller. But unless I’m hammering the thing with sustained writes, am I likely to notice the difference?

      • Waco
      • 5 months ago

      The answer to that is no, for basically any non-garbage NVMe drive.

    • boidsonly
    • 6 months ago

    I bought the 512gb model and am happy with its performance. Not the cheapest NVMe out there but the speed is good enough for the price.

    • Stochastic
    • 6 months ago

    Not bad, although for more casual users it’s hard to beat the bang-for-the-buck you get from the Intel 660p (currently $110 for 1 TB on Newegg).

    • Alistair
    • 6 months ago

    Any chance of a quick test of loading challenge du jour – Anthem?

    🙂

      • cynan
      • 6 months ago

      If going to test loading times, might as well use a game that people are still playing?

    • DPete27
    • 6 months ago

    I bet once the SM2262EN is paired with 96 layer vNAND like the 970Evo+, we’ll see a new king. I find it odd that the SX8200Pro and HP EX950 didn’t make it with 96 layer vNAND. These products seem like stop-gap.

    • Voldenuit
    • 6 months ago

    What’s the size of the SLC write cache? Is it in the drive specs? From the IOMeter graphs, it looks to be somewhere around 80 GB, does that sound right? Does the size of the SLC cache depend on how much free space is on the drive and/or depend on wear leveling algorithms?

    I doubt anyone outside of dedicated servers is writing 80-100 GB at a time, but I am curious for curiosity’s sake.

      • Goty
      • 6 months ago

      Anandtech has some comment on that: [url<]https://www.anandtech.com/show/13759/comparing-adata-sx8200-pro-vs-hp-ex950/2[/url<] Interestingly, their test suite shows this drive and its HP cousin in a much less flattering light.

        • DPete27
        • 6 months ago

        That was a very informative article. Thanks for mentioning it!!

          • Goty
          • 6 months ago

          Anandtech has some pretty good storage reviews, and they might be the most thorough around now that Alan is no longer at PCPer.

            • DPete27
            • 6 months ago

            Agreed.

    • enixenigma
    • 6 months ago

    [quote<]They both lean on pseudo-SLC caching to reach their peak speeds, both omit hardware encryption acceleration, both have the same five-year warranty, [b<]and both are rated for the same number of total bytes written—640 GB for the 1 TB flavor[/b<].[/quote<] Not sure I follow.

      • shaq_mobile
      • 6 months ago

      Does that mean you should avoid filling up your SSD? 🙂

      • weaktoss
      • 6 months ago

      Whoops, should have read “640 TB.” Fixed, thanks!

        • Chrispy_
        • 6 months ago

        That really hammers home the reduced write/erase cycles of TLC NAND these days.

        Back when NAND cost 10-20x what it does now, SLC had an endurance of 10K cycles, MLC had 5K cycles, and early planar TLC had 3K cycles.

        When a manufacturer is effectively stating 640TB endurance per TB of NAND, we have dropped from 3K cycles to 640 cycles.

        I’m assuming that’s a combination of multiple factors:
        [list<][*<]Race-to-the-bottom means that NAND is now lower-quality and has more defects than ever before[/*<][*<]TLC NAND performance needs more error correction and background activity than ever before to hide the performance issues, so write amplification has gone through the roof[/*<][*<]Increasing layers for more capacity per die may have a detrimental effect on the endurance.[/*<][*<]Now we're at commodity pricing for NAND, the lack of profit means that vendors don't want to offer any kind of promises that will invoke RMAs. For a 5 year warranty, the low 'TB-written' value will likely excempt them from failures on heavily-used drives (even consumer-level 1080p video editing will chew through that 640TB in no time!)[/*<][/list<] I thought 3D NAND was supposed to allow capacity increases without problematic and endurance-harming process node shrinks, but clearly the fabs and manufacturers are getting greedy and reverting back to more capacity/die with process shrinks - to the detriment of both access latency and endurance. You'd think they'd learn from the issues caused by previous planar node shrinks, but as long as these things operate acceptably for the warranty period, for an acceptable number of casual users, they don't care how horrendous the problems after that will be for their heavy-use customers. Today it's 640 cycles. Next year it may be just 250 cycles. At what point will degredation and failures cause the consumers to start pushing back? 😛

          • Waco
          • 6 months ago

          This isn’t really anything new – consumer drives have almost always been rated between 500-1000 overwrites almost regardless of the type of NAND inside.

          3D TLC NAND is a lot closer to 2D MLC in endurance than planar TLC NAND in practice. 3D MLC is impressively endurant, almost to the level of OG SLC.

      • UberGerbil
      • 6 months ago

      “640 [i<]whatevers[/i<] should be enough for anybody." -- Bill Gates, probably.

    • willmore
    • 6 months ago

    Another plea to use consistant scaling on all the graphs so that meaningful comparisons can be made when clicking amongst the various drives.

    Great review. Good to see Samsung getting some competition. That can only encourage them to do better and provides more good performing drives for consumers. Well done Adata.

      • weaktoss
      • 6 months ago

      Which graphs exhibit the inconsistency you’re talking about? The first set of IOMeter scaling graphs intentionally use a different scale per drive, since the point of them is to highlight a single drive’s changes in behavior at different queue depths. And that’s why we put in the second set of scaling graphs, for context against other drives. It seems to me that all the other graphs make it easy to compare drives against each other.

        • willmore
        • 6 months ago

        Those IOMeter graphs on page 3 are the ones I’m refering to. If each drive was to be evaluated alone, then having each on its own scale makes sense, but the purpose of having them all in the review of a *different* drive is to compare them to the drive in question. Therefore they should all be on the same scale. The shape doesn’t matter in this context because it’s not looking at the characteristics of *one* drive, but the relative differences between two (or more) drives.

        The graphs which compare different drives “IOMeter Transaction rate Database Total” graph has insufficent contrast amongst the colors. At least for people with deuteropia. Since I’m colorblind that way, I don’t tend to rely on that specific graph.

    • Usacomp2k3
    • 6 months ago

    Is there any chance we can get a Performance per $$ graph without the synthetic and unrealistic data points? Ideally just load times and 1T read/write/copy performance?

Pin It on Pinterest

Share This